Antibiotic therapy against Pseudomonas aeruginosa in cystic fibrosis: a European consensus

Eur Respir J 2000; 16: 749±767 Printed in UK ± all rights reserved Copyright #ERS Journals Ltd 2000 European Respiratory Journal ISSN 0903-1936 CONSENSUS STATEMENT Antibiotic therapy against Pseudomonas aeruginosa in cystic fibrosis: a European consensus G. DoÈring*, S.P. Conway**, H.G.M. Heijerman #, M.E. Hodson ##, N. Hùiby 1, A. Smyth +, D.J. Touw ++, for the Consensus Committee Antibiotic therapy against Pseudomonas aeruginosa in cystic fibrosis: a European consensus. G. DoÈring, S.P. Conway, H.G.M. Heijerman, M.E. Hodson, N. Hùiby, A. Smyth, D.J. Touw, for the Consensus Committee. #ERS Journals Ltd 2000. ABSTRACT: Cystic fibrosis (CF) is the most common lethal hereditary disorder with autosomal recessive heredity in caucasians. The majority of CF patients suffer from chronic respiratory infection with the opportunistic bacterial pathogen Pseudomonas aeruginosa. No consensus among clinicians has been reached so far concerning antibiotic treatment against P. aeruginosa in CF patients. Consensus answers to 24 important questions in this context, based on current evidence, are presented, given by a panel of 34 European experts. Questions addressed and answered are: The diagnosis of P. aeruginosa lung colonization in CF; The impact of P. aeruginosa on the clinical state of CF patients; The assessment of P. aeruginosa susceptibility against antibiotics and the importance of these results for the clinician; The use of monotherapy versus combination therapy; The development of microbial resistance; The achievement of optimal airway concentrations; The effects of subinhibitory concentrations of antibiotics on P. aeruginosa; Statements on the pharmacokinetics of antibiotics in CF patients; Recommendations for doses and dosing intervals and length of treatment regimens; and Toxic side effects due to repeated antibiotic therapy was addressed. The expert panel answered further questions on the use of fluoroquinolones in children with CF, on the administration of nebulized antibiotics and whether prevention of P. aeruginosa lung colonization is possible in CF using antibiotic therapy. Problems of antibiotic therapy at home and in the hospital were addressed, a consensus statement on regular maintenance treatment, or treatment on demand, was given and different routes of administration of antibiotics were recommended for different clinical situations. Finally, the factors which determine the choice of the antibiotic, the dosage, and the duration of the treatment in cystic fibrosis patients were addressed and the design of future antibiotic studies in the context of Pseudomonas aeruginosa lung infection in cystic fibrosis patients were recommended. Eur Respir J 2000; 16: 749±767. *Hygiene-Institut, University of TuÈbingen, TuÈbingen, Germany, **St Jame's & Seacroft Hospital, Regional Adult Cystic Fibrosis Unit, Leeds, UK, # Ziekenhuis Leyenburg, Den Haag, the Netherlands, ## Royal Brompton Hospital, The Cardiothorcic Institute, London, UK, " Laboratory Centre, Klinisk Mikrobiologisk afdeling, Copenhagen, Denmark, + Dept of Paediatrics, Nottingham City Hospital, Nottingham, UK, + +Academic Hospital Vrije Universiteit, Dept of Pharmacy, Amsterdam, the Netherlands. Correspondence: G. DoÈring, Dept of General and Environmental Hygiene, Hygiene-Institute, University of TuÈbingen, Wilhelmstrasse 31, D-72074 TuÈbingen, Germany. Fax: 49 7071293011 Keywords: Antibiotic therapy, consensus, cystic fibrosis, Pseudomonas aeruginosa Received: May 2 2000 Accepted after revision June 20 2000 This document is the result of an European Consensus Conference which took place in Artimino, Tuscany, Italy, in November 1999 involving 34 experts on antibiotic therapy against Pseudomonas aeruginosa in cystic fibrosis patients, organized by the European Cystic Fibrosis Society, and sponsored by PathoGenesis (major sponsor), Astra-Zeneca, Aventis, Bayer, Glaxo- Wellcome, Lilly and Pharmax. The purpose of the conference was to develop a consensus document on antibiotic treatment against P. aeruginosa in cystic fibrosis patients based on current evidence. For editorial comments see page 583. Cystic fibrosis (CF) is a worldwide disease occurring in virtually all ethnic groups. In Caucasians it is the most common lethal hereditary disorder with autosomal recessive heredity [1]. About 1 in 25 are heterozygous carriers, while the incidence of clinical disease is ~1 in 2,500 live births [1], with a prevalence of ~50,000 recognized patients worldwide. Disease is caused by mutations in a single gene of chromosome 7, which encodes the CF transmembrane conductance regulator (CFTR) [2]. The CFTR protein is a membrane bound cyclic adenosine monophosphate (camp)-regulated chloride channel which is thought to regulate other cell membrane ion channels [3]. More than 800 different mutations have been identified so far, however, a deletion of phenylalanine in the amino acid position 508 is present in ~66% of all CF patients worldwide. CFTR mutations affect epithelial ion and water transport mainly in cells in the respiratory, gastrointestinal, hepatobiliary and reproductive tracts. In CF airways diminished chloride and water secretion leads to viscous secretions and impaired mucociliary clearance [3]. A hallmark of CF is chronic respiratory infection, which may start very early in the life of these patients [4±6]. Since the early description of CF, pulmonary infection has been recognized as having the greatest role in morbidity and mortality leading to premature death in 90% of patients [5]. Pulmonary infections in CF are unique because of the underlying basic defect, the restriction of infection to the pulmonary tract and the chronic nature of infection with progressive tissue damage due to chronic inflammation.

750 G. DOÈ RING ET AL. Pseudomonas aeruginosa, a Gram-negative bacterium, which also causes infections in patients with severe burns, cancer, and paraplegia [7], is the most frequently described opportunistic pathogen in CF patients. It is ubiquitous and is found in many water reservoirs in the inanimate and human environment [8]. Water reservoirs polluted by humans or animals, such as with sewage, are the most frequent environmental sources of P. aeruginosa. P. aeruginosa is seldom and transiently isolated from healthy humans. P. aeruginosa is the most frequent pathogen isolated from CF sputum or bronchoalveolar lavage samples of CF patients of all age groups [5, 9, 10]. Many children with CF develop respiratory tract infection early in their lives. Epidemiological data from the Cystic Fibrosis Foundation, USA, including ~20,000 CF-patients, revealed that 29.8% in the age groups of 2±5 yrs and 81.3% in the age group of 26±30 yrs are infected with P. aeruginosa [5]. Similar values are obtained in European countries [10]. An overall incidence of P. aeruginosa infections in CF is estimated at 8%. Epidemiological studies using different typing methods for P. aeruginosa suggest that transmission of P. aeruginosa to CF patients may occur by a direct patient-topatient contact. In addition, transmission of P. aeruginosa to CF patients may occur via contaminated environmental reservoirs [8]. CF centres have separated CF patients with and without P. aeruginosa infection in order to limit cross-infection [11]. In addition, hygienic measures to decontaminate environmental reservoirs of P. aeruginosa including nebulizers [12] and other medical equipment, sinks, toilets and dental tubings, have been recommended. Additionally hand disinfection for CF patients and hospital personnel has been stressed [8]. Vaccination against P. aeruginosa in CF patients has been studied [13]. Whereas in most trials vaccines do not appear to prevent lung infection in CF patients [14, 15], administration of a polysaccharide-exotoxin A conjugate vaccine may have some clinical efficacy [16]. Currently, a multi-centre trial with a bivalent P. aeruginosa flagella vaccine including >500 CF patients in Europe is in progress [17]. Pathogenesis of the chronic Pseudomonas aeruginosa lung infection in cystic fibrosis patients The exact route(s) of entry of the bacteria into the lower airways of CF patients is not known. It is generally believed that the pathogen reaches the lower airways through mouth or nose passage. The reason for the peculiar interaction of P. aeruginosa with the CF lung epithelium is unclear at present, although several hypotheses have been offered. Binding of P. aeruginosa to components of the mucus layer on top of the cilia and failure to clear thickened mucus from airway surfaces is likely [18]. P. aeruginosa colonization of CF airways triggers an inflammatory response with concomitant release of a number of cytokines, such as interleukin-8, which is chemotactic to polymorphonuclear leukocytes (neutrophils). Whereas mostly nonmucoid P. aeruginosa is cultured in vitro from oropharyngeal CF specimens soon after onset of infection, mucoid P. aeruginosa, characterized by the formation of exopolysaccharide (alginate)-coated microcolonies (biofilm mode of growth), is nearly uniformly present in chronic infection [9]. Mucoid P. aeruginosa is more resistant against phagocytosis [19] and may also restrict the penetration of antibiotics. Mucoid P. aeruginosa is often nonmotile and this phenotype persists in a nutrient rich, mostly anaerobic environment. The pathogenesis of chronic P. aeruginosa lung infection in CF has been classified as a type III hypersensitivity reaction, characterized by the production of specific antibodies against a large number of bacterial antigens, immune complexes and large numbers of neutrophils [9, 20]. The decaying neutrophils form large areas of pus around the persisting bacteria, which may lead to total obstruction of CF airways. The release of high extracellular concentrations of lysosomal serine proteinases far exceeds the endogenous antiproteinase capacity. Progressive proteolytic impairment of multiple defense pathways in addition to endobronchial tissue destruction are thought to be responsible for the significantly reduced life expectancy in CF patients [20]. Diagnosis of Pseudomonas aeruginosa lung infection and its clinical course in cystic fibrosis patients P. aeruginosa lung infection in CF patients is generally diagnosed by taking sputum, hypopharyngeal or endolaryngeal suctions or deep throat culture, preferably after lung physiotherapy or alternatively after hypertonic (3%) saline inhalation. P. aeruginosa grows well on standard laboratory media such as 5% sheep blood agar or chocolate agar [21, 22]. Selective media, such as cetrimide agar, facilitate the isolation of P. aeruginosa from other bacterial species. Plates are incubated at 358C in a 5% CO 2 atmosphere for 24 h. P. aeruginosa is easily identified by characteristic colony morphology and standard microbiological methods [21, 22]. P. aeruginosa isolates may differ concerning their morphotypes including mucoid, smooth, rough, dwarf and small colony varieties, and morphological varieties from a single patient may differ in antibiotic susceptibility patterns [22, 23]. However, different morphotypes mostly belong to a single genotype. The time required to identify P. aeruginosa and assess its antibiotic susceptibility pattern from a sputum culture is ~3±4 days. However, small colony variants may only grow after 48 h agar cultivation and thus may be overlooked in routine diagnostics [24]. Gram-stained smears of patient materials are used to detect neutrophils and squamous epithelial cells besides micro-organisms and will indicate the adequacy of the sputum sample. In non-expectorating patients oropharyngeal cultures yielding P. aeruginosa are highly predictive [25]. However, false-negative results from throat cultures may occur making additional diagnostic tools necessary. Sensitive serological tests for P. aeruginosa antigens such as crossed immunoelectrophoresis, radioimmunoassay (RIA) and enzyme-linked immunosorbent assay (ELISA) have been especially useful in this context [26]. The presence of P. aeruginosa in throat culture and negative P. aeruginosa antibody titres may help to distinguish between early colonization and chronic infection. P. aeruginosa colonization/infection may start in the first year of life of

ANTIBIOTIC THERAPY AGAINST P. AERUGINOSA IN CF 751 a CF patient [3, 4, 6, 10, 27] and it is believed that early therapeutic intervention will delay progressive lung damage caused by P. aeruginosa or even prevent or delay the onset of chronic infection. It has been demonstrated that clinical progression of lung disease with deterioration of forced expiratory volume in one second (FEV1) occurred despite the disappearance of P. aeruginosa from sputum specimens, suggesting that lung disease can also progress with other bacteria [28]. Additionally, epidemiological data show that infection with P. aeruginosa is confounded with the degree of pulmonary impairment which already exists [29] and that P. aeruginosa, although associated with some decrease in FEV1, does not cause sudden deterioration in CF patients [30]. Apparently, some patients can tolerate P. aeruginosa colonization for years without decline of lung function. They are frequently, though not necessarily, characterized by harbouring nonmucoid P. aeruginosa strains, absent or low antibody titers to the organisms and absent or little sputum production. However, in the majority of CF patients producing sputum and carrying mucoid P. aeruginosa, the infection causes an immediate and more rapid reduction of lung function than in CF patients without P. aeruginosa infection [9]. This leads to the hypothesis that uncontrolled progression of infection will result in progressively severe lung damage, respiratory failure and death. Indirect evidence for the latter hypothesis stems from the fact that sputum volumes increase significantly after onset of P. aeruginosa infection. Higher sputum volumes are correlated with a higher degree of inflammation, higher neutrophil numbers and released serine proteinases and hence higher degree of lung obstruction and destruction. Not all clinicians, however, agree to this description of the pathophysiology of the P. aeruginosa lung infection, as pulmonary exacerbations from other causes may lead to similar clinical changes. The use of antibiotics for P. aeruginosa lung infection in CF patients has been questioned by only a few investigators. When P. aeruginosa-infected CF patients were treated for acute exacerbations with antibiotics, the shortterm clinical outcome was comparable to the placebo treated group [31]. Based on unchanged sputum colony counts during or after intravenous antibiotic therapy, it was concluded that antipseudomonal therapy in CF patients requires more critical bacteriological and clinical evaluation [32]. However, several other studies investigating various antibiotics for P. aeruginosa in CF patients found a positive effect on clinical condition, pulmonary function, P. aeruginosa colony counts in sputum, inflammatory parameters, quality of life and nutritional status of the patients. Epidemiological data showed that lung function is the best predictor of survival [29] and antibiotic therapy has lead to improvement in lung function in many studies [33, 34]. The observation that the degree of improvement in lung function correlated with the degree of reduction of P. aeruginosa density in sputum specimens [34] suggests that antibiotic therapy is directly responsible for lung function improvement. For long-term effect, however, the influence of anti-p. aeruginosa treatment is difficult to evaluate since no controlled studies have been performed and such long-term studies would be difficult to carry out without well defined therapeutic strategies and due to complexity of the management of the patient and a limited patient population. CF is a multi-faceted disease and many different treatment strategies have been implemented during the last 40 yrs. Therefore, it is difficult to prove that antibiotic treatment alone is responsible for the dramatic positive change in survival rates of CF patients in some countries. Besides antibiotics, improved mucolytic therapy, airway physiotherapy, improved nutritional strategies including the use of pancreatic enzyme preparations have contributed to the increased life expectancy in CF patients. The median survival age of CF patients in the USA increased from 14 yrs in 1969 to 31.3 yrs in 1996 [5]. In Denmark, the probability of surviving 40 yrs for a diagnosed CF patient was 83.3% in 1995 [35]. This improved survival has been correlated with an early anti-p. aeruginosa treatment regimen in Denmark [35]. Antibiotic resistance of Pseudomonas aeruginosa in cystic fibrosis Traditionally, many decisions to treat P. aeruginosa lung infections in CF patients are empirical and are based in part on patient's age, severity of disease and past experience rather than on P. aeruginosa susceptibility patterns against antibiotics. Bacterial susceptibility (or resistance) patterns against antibiotics are assessed in vitro using the relatively simple disk diffusion method, which is regarded to give qualitative results according to established break-points, or the quantitative dilution methods (agar dilution, microdilution broth, macrodilution broth). By using an automated broth microdilution method, synergistic effects of combinations of different antibiotics can be demonstrated which may help to guide antibiotic therapy regimes in CF patients [36]. For sputa containing susceptible isolates, testing the mixed culture may provide adequate susceptibility data with significant laboratory time and cost savings [23]. However, for sputa with resistant strains, the traditional method of testing isolated morphotypes may be more appropriate [23]. Several problems arise when P. aeruginosa susceptibility patterns against antibiotics are used for clinical treatment in CF patients: 1) multiple morphotypes with different antibiotic susceptibilities are often isolated from a single sputum sample [22, 23]; 2) the conventional definition of antibiotic resistance, which is based on the comparison of serum drug levels and minimal inhibitory concentrations (MIC) in vitro does not apply to CF lung infections when treated with inhaled antibiotics [37]; 3) The biofilm mode of growth of P. aeruginosa requires 100±1,000 times the concentration of a certain antibiotic to be effective compared to its nonmucoid variant which is normally tested [38]. Nevertheless, for several antibiotics tested, nonmucoid isolates were more resistant than mucoid isolates [39]. Whether CF patients should be treated over prolonged periods of time with antibiotics to which P. aeruginosa is resistant, according to definitions used in Europe or North America, remains unclear at present. Actual data on P. aeruginosa susceptibility patterns against antibiotics are missing in Europe but may be retrieved from CF databases.

752 G. DOÈ RING ET AL. When antibiotic therapy had been started before susceptibility patterns had been available, the therapy has to be adjusted accordingly. Chronic P. aeruginosa infections in CF necessitates many antibiotic therapy courses which carry the risk that resistance towards a given drug increases more or less rapidly. Consequently, P. aeruginosa often display higher resistance to older drugs which have been used more frequently compared to newly developed drugs [40±42]. Also trials with nebulized antibiotics in CF patients, resulted in the emergence of resistant P. aeruginosa isolates in the sputum during and after treatment with great variations [33, 43±47]. For instance, a study of inhaled tobramycin showed that the percentage of patients harbouring P. aeruginosa strains with a tobramycin MIC >8 mg.ml -1 increased from 29±73% after 3 months of continuous use of 600 mg three times daily [47]. Therefore, studies using an intermittent regimen of inhaled tobramycin have been conducted [33, 45, 46]. With this regime, the percentage of patients harbouring P. aeruginosa with a tobramycin MIC >16 mg.ml -1 increased from 13± 25% only. In a subsequent open tobramycin study over 18 months using a daily dose of 600 mg, the increase of the MIC of tobramycin had no negative effect on lung function of the patients [48]. The emergence of resistant micro-organisms has been reported to be transient and reverted to susceptible over time, when antibiotic selective pressure is removed, a phenomenon which was called "adaptive resistance" [49]. The mechanism which leads to adaptive resistance may be based on a slower growth rate of resistant colonial forms of P. aeruginosa compared to susceptible ones. Generally, nonmucoid P. aeruginosa is more resistant to antibiotics than mucoid P. aeruginosa [39]. Consequently, susceptible organisms will overgrow resistant forms after the end of the antibiotic course. Frequent changing from one antipseudomonal antibiotic to another may be an option to prevent antibiotic resistant P. aeruginosa. The emergence of resistant micro-organisms does not necessarily lead to a poor response to repeated treatment [41, 44, 47, 50, 51]. Nevertheless, the emergence of multiresistant P. aeruginosa strains from CF patients is well known. Current epidemiological data on the prevalence of antibiotic resistant P. aeruginosa infecting CF patients in European countries other than Italy [52] and Denmark [53] are missing. Depending on the mode of action of a given antibiotic, more or less rapid development of resistance occurs [54] especially with monotherapy, when the drug has to be given for prolonged periods of time. Indeed, a high development of ceftazidime-resistant strains of P. aeruginosa was noted in a small population of CF patients who received ceftazidime monotherapy over a 5-yr period [55, 56]. This would imply that antibiotic monotherapy is inferior to combination therapy with P. aeruginosaspecific antibiotics displaying different mode of actions. For example, b-lactam antibiotics such as the penicillins and cephalosporins, interfere with bacterial cell wall biosynthesis, novobiocin and the quinolones inhibit deoxyribonucleic (DNA) synthesis by acting on topoisomerases and rifampin inhibits DNA-directed ribonucleic (RNA) synthesis. In contrast, aminoglycosides as well as chloramphenicol, erythromycin, clindamycin and the tetracyclines inhibit protein synthesis by binding to bacterial ribosomal subunits or inhibiting ribosomal enzymes [57]. However, when intravenously administered ticarcillin and tobramycin was compared to ceftazidime, development of resistance to multiple antibiotics was significantly lower with ceftazidime alone compared to the ticarcillin/ tobramycin group [58]. Other studies revealed that development of resistant strains is not prevented by the use of a combination of tobramycin and b-lactam antibiotics [59, 60]. In addition, emergence of resistant micro-organisms during aerosolization of colistin alone seems to be rare, although published evidence is lacking. Nevertheless, combination therapy is widely used in CF patients. Selection of intrinsically resistant pathogens after antibiotic therapy in CF patients may occur, yet no evidence of increased selection for intrinsically tobramycin-resistant bacterial pathogens, including Burkholdia cepacia, Stenotrophomonas, and Alcaligenes xylosoxidans was reported when intermittent inhaled tobramycin therapy was compared with standard antimicrobial therapy [37]. However, there was increased isolation of Candida albicans and Aspergillus spp. in the treatment group compared with the placebo group [37]. Such a development was not observed in a previous study using aerosolized colistin in CF [43]. The clinical significance of these findings is unclear, since long-term data on CF patients who have received inhaled tobramycin for several years are missing. Penetration of antibiotics into respiratory secretions and pharmacokinetics of antibiotics in cystic fibrosis patients Antibiotics are believed to enter the bronchial secretions from the blood by simple passive diffusion along a concentration gradient [61]. Aminoglycosides diffuse poorly across lipid membranes and into bronchial secretions and their distribution is mainly restricted to the extracellular fluid. Although tobramycin has the highest penetration into bronchial secretions of aminoglycosides studied [62, 63], relatively high dosages are needed when the drug is given parenterally. Therefore, it is generally recommended to administer aminoglycosides by inhalation for maintenance therapy. Intravenous aminoglycoside doses have to be adjusted according to the serum levels of the patient because of toxicity and efficacy of the drugs and therefore may change from patient to patient. Doses can be adjusted based on a mathematical model [64], which, however, is not available in many centres. A practical approach is to determine serum trough and peak levels of aminoglycosides. Data on airway concentrations of aminoglycosides [47, 65] as well as on other antibiotics [66, 67] are sparse and depend on the mode of delivery [67]. Once daily dosing of amikacin and tobramycin achieves higher airway concentrations than multiple daily dosing [68]. The general applicability of once daily dosing is, however, pending. When the drugs are used via the inhaled route during acute exacerbations clinical experience suggests low efficacy and therefore intravenous use is preferred [69]. Further studies which compare intravenous and inhaled administration of antibiotics during acute exacerbations are necessary.

ANTIBIOTIC THERAPY AGAINST P. AERUGINOSA IN CF 753 Table 1. ± Recommended dosages for antibacterial agents in the management of P. aeruginosa lung infections in cystic fibrosis patients Antibiotics Route of administration Dose mg.kg -1.day -1 Administrations per day n Maximum daily dose g Amikacin* i.v. 30 2 - Aztreonam i.v. 150 4 8 i.v. 100 continuously 8 Cefepime i.v. 100±150 2±3 6 Ceftazidime i.v. 150±250 3±4 12 Ceftazidime i.v. 100±150 continuously 12 Ciprofloxacin p.o. 30 2±3 1.5±2.25 Colistin Inhaled 80±160 # 1±2 0.320 # Sulphomethate i.v. 160 # 3 0.48 Imipenem/cilastatin i.v. 50±100 3±4 4 Meropenem i.v. 60±120 3 6 i.v. 60 continuously 3 Netilmicin* i.v. 10 2 - Ticarcillin i.v. 500±750 4 30 Tobramycin* i.v. 10 2 - Inhaled 150±300 # 1±2 0.6 *: dose based on measurements of serum concentrations; # : absolute dose (dependent on age and situation). These recommendations may not be the approved dosing recommendations by regulatory authorities in different countries. Penicillin and cephalosporins do not enter bronchial secretions well and bronchial secretions and sputum concentrations usually range 3±15% of plasma levels [70]. Based on a minimum inhibition concentration inhibiting 90% of tested strains (MIC90) for ceftazime and P. aeruginosa of 8 mg.ml -1, the amount of drug to be administered parenterally every 6 h would be 300 mg.kg -1 [70]. However, when administered continuously, a much lower dose can be used. Penicillins and cephalosporins are generally administered intravenously. For classes of antibiotics other than aminoglycosides given intravenously, monitoring of serum levels is not required because of their wide therapeutic index and recommended doses as given in table 1 should be used. Administration of ciprofloxacin orally has resulted in sputum concentrations ranging from 46±90% of those obtained in serum [64, 71]. flouroquinolones are given either intravenously or orally. The latter route is preferred because of good absorption. The intravenous route is much more expensive and less convenient for the patient. Colistin given by inhalation is widely used for maintenance therapy and for eradication of early colonization/ infection with P. aeruginosa. Due to its chemical nature, a product which upon hydrolysis releases the active drug, a cycle polypeptide, analysis of colistan airway concentrations are impossible to preform and therefore the optimum levels are not known. Efficiency has been shown at doses ranging from 1 million units twice daily [43] to 2 million units three times daily [72]. Colistin has been used with efficiency by the parenteral route in patients with multiresistant P. aeruginosa infection [73, 74]. In addition to penetration into bronchial secretions, plug formation may present another problem for antibiotic action. CF plugs are mainly composed of negatively charged glycoproteins and DNA. Other macromolecules, and positively charged aminoglycosides. such as tobramycin, may be bound to these compounds [75±78]. Because of these reasons appropriate airway clearance should be performed before inhalation of aminoglycosides or colistin. No data on flouroquinolones binding to charged macromolecules are available. Ceftazidime binding to sputum is negligible [78]. A further obstacle for antibiotics to reach the P. aeruginosa outer cell membrane in CF plugs is the mucoid exopolysaccgaride (alginate) matrix [79]. In vitro models have been developed to evaluate the penetration of ceftazidime into plugs [64]. Taken together, bactericidal concentrations of antibiotics in CF plugs are difficult to reach and it has to be assumed that often only subinhibitory concentrations can be obtained. The same is true for poorly ventilated areas. Subinhibitory concentrations of antibiotics may positively affect the clinical condition and the lung function of CF patients [80±84]. For instance, ciprofloxicin [81] and gentamicin [83] inhibit exotoxin production of P. aeruginosa, which may decrease inflammation in CF due to decreased immune complex formation. Tobramycin and gentamicin inactivate oxygen radicals by binding to amino groups [80] and ticarcillin and ceftazidime may protect a 1 -antitrypsin from oxygen radical damage [80]. Whether such modifications will lead to inactivation of antibiotics is unknown. However, subinhibitory concentrations of antibiotics may change the mutation rate of P. aeruginosa facilitating the development of resistant variants (adaptive mutation). They may also trigger adaptive resistance [49]. The pharmacokinetics of antibiotics in CF patients have been extensively studied [64, 76, 85±93]. CF patients often reveal different pharmacokinetics particularly for aminoglycosides and b-lactam-antibiotics when compared with normal individuals. The volume of distribution per kg body weight is often increased and the elimination half-life decreased [90]. The higher body clearance due to increased renal and nonrenal elimination [76, 88] demands that higher antibiotic doses are administered more frequently [87, 93]. However, differences between CF patients and non-cf individuals in this context do not seem to be as large as used to be thought and may have been exaggerated by the lack of properly matched controls for

754 G. DOÈ RING ET AL. clinical studies [69]. The pharmacokinetics of ciprofloxacin are not altered in CF patients compared to non-cf individuals [94±96]. However, due to the basic defect, delayed absorption of oral antibiotics including ciprofloxaxin has been reported [91]. Intravenous antipseudomonal antibiotic therapy in cystic fibrosis Mostly due to the endobronchial location of the P. aeruginosa infection in CF patients and the inaccessibility of mucoid P. aeruginosa in plugs, high doses of antibiotics have been recommended in CF patients (table 1). Nevertheless, optimal doses of various intravenously administered antibiotics in CF are debated. The initial doses of aminoglycosides may be calculated based on the body surface area [76] or body weight [64, 90, 93]. A theoretical calculation of the dosing of ceftazidime demonstrates that the previously recommended doses of <200 mg.kg -1 [97] may still be too low for treating CF patients [70] when given intermittently. Meropenem was given at a dose of 40 mg.kg -1 [98] three times daily. Intravenously administered aminoglycosides are usually given three times daily. Initial tobramycin dosages of ~10 mg.kg -1.day -1 and more have been recommended [22, 64, 99, 100]. An adjustment in dose interval based on the serum tobramycin level has a significant effect on pulmonary function [93]. Due to the subinhibitory MIC effect [101], the bactericidal concentration-dependent killing and adaptive resistance, aminoglycosides might be administered in higher doses less frequently [102]. Indeed, 12 hourly high peak aminoglycoside dosing has been shown to be less toxic and equally efficacious as 8 hourly dosing [103]. Similarly, a study comparing once daily with thrice daily intravenous tobramycin suggested that once daily tobramycin combined with three injections of ceftazidime is safe and effective in CF [68]. To fully answer the question on efficacy and toxicity in this context, a multicentre study is currently being carried out in the UK. (A. Smyth, Dept of Paediatrics, Nottingham City Hospital, Nottingham, UK, personal communication). Since piperacillin has been repeatedly described to induce a serumsickness-like symptom [104±106], caution is needed when using this drug in high concentrations. Intravenous antibiotic therapy is generally scheduled for ~2 weeks but may also be prolonged for another week [99]. Courses with intravenously administered antibiotics have been shown, by most investigators, to be clinically effective. The effectiveness of the various penicillins seems to be similar [42]. From the third generation cephalosporins, ceftazidime particularly is effective against P. aeruginosa [85, 97]. Similarly, no differences were found in effects on clinical condition, pulmonary function, or P. aeruginosa sputum counts when different aminoglycosides were compared with each other such as tobramycin with gentamicin or with netilmicin [42]. From a clinical viewpoint and depending on the local antimicrobial resistance patterns encountered most frequently, gentamicin appears the less desirable option. Combination therapy with antibiotics has been favoured by several investigators because it may slow down drug resistance and may result in synergy. Often, a combination of a penicillin and an aminoglycoside is given. Gentamicin/carbenicillin, tobramycin/ticarcillin or amikacin/ceftazidime combinations have been widely used. Probenicid may be taken orally to enhance blood levels of some penicillins. Cefsulodin, piperacillin, imipenem, fosfomycin or aztreonam have been combined with tobramycin or amikacin [42]. Combinations of aminoglycosides with a cephalosporin such as ceftazidime or cefsulodin seem to be superior in their effect on pulmonary function, clinical situation and drug resistance compared to combinations with penicillins. Also the combination of antipseudomonal antibiotics with antibiotics of less efficacy against this organism may lead to beneficial results. For example, in vitro the combination of high concentrations of ceftazidime and gentamicin with rifampicin, which has little activity against the strains, led to the elimination of P. aeruginosa from biofilms [107]. Whether macrolides such as azithromycin [108] are effective has to be investigated in future multicentre studies. In vitro studies showed an inhibitory activity of macrolides on the production of alginate and other virulence factors of P. aeruginosa [109]. Nevertheless, in some studies, monotherapy was superior to combination therapy. For example, when ticarcillin and tobramycin was compared to ceftazidime, clinical outcome was comparable, yet ceftazidime was more efficient to reduce P. aeruginosa colony counts in sputum specimens [58]. Similarly, ceftazidime alone showed better effectiveness than tobramycin and carbenicillin and better follow-up lung function values [110, 111]. When azlocillin alone was compared to azlocillin and tobramycin [60] at the end of the treatment no significant difference was seen between the two groups in clinical evaluation, sputum DNA concentration, forced vital capacity, FEV1, or peak expiratory flow rate. However, sputum P. aeruginosa density decreased more with combination therapy and time to readmission for a new pulmonary exacerbation was significantly longer in the group receiving combination therapy [60]. Thus, little difference existed between antibiotics or their combinations concerning clinical condition, bacterial counts in sputum, leukocytes and chest radiographic findings [42]. Clinical studies evaluating the rational use of the combination of intravenous and aerosolized antibiotics in acute exacerbations are scarce and conclude that this combination therapy is not superior compared to intravenously administered therapy alone [69, 112±115]. CF patients are treated with antibiotics for most of their lives. Thus, any new or unusual symptom must be viewed with great suspicion as a possible drug side effect. The potential toxicities of aminoglycosides are similar and largely confined to the vestibular-auditory and renal systems [116] due to accumulation of the drug in cells where they bind to the ribosomes. Typical manifestations of ototoxicity such as loss of hair cells and degeneration of ganglion cells have been reported in CF patients [117, 118]. This may lead to high tone deafness and hypokalaemia or hypomagnesaemia [119]. Acute vestibular toxicity may be seen if the drug is given rapidly [102] and a long infusion is preferable. Some studies suggest tobramycin may be less nephrotoxic than gentamicin [119]. Administration of 15 mg.kg -1.day -1 of tobramycin [102] and netilmicin [120] has been associated with symptoms of dizziness after the antibiotic infusion. However, surprisingly few side-effects have been reported in high dose

ANTIBIOTIC THERAPY AGAINST P. AERUGINOSA IN CF 755 aminoglycoside therapy [89, 121, 122]. Nevertheless, it is necessary to monitor ototoxicity and renal function and adjust the dosage interval accordingly, early in therapy [64]. When meropenem was used in CF patients only transient elevations of serum transaminases have been reported and none of the patients experienced nausea and vomiting, even when meropenem was administered as a bolus injection [123]. Concerns have been raised on the potential toxicity of intravenous colistin. However, two studies reported a low frequency of nephrotoxicity [73, 74]. The UK is the only known European country where colistin is given intravenously in CF patients. Whether other compounds present in the formulation of antibiotics such as phenol in the tobramycin preparation lead to toxicity is unknown. No written reports on side effects in this context have been collected by one of the manufacturers of tobramycin worldwide (J. Biclher, Lilly Germany, Bad Hamburg, Germany, personal communication). Penicillins contain large amounts of sodium, and caution is required in patients at risk of cor pulmonale. The addition of vitamin K to the daily regimen of patients on chronic or highdosage antibiotic therapy has been recommended [22]. More frequent are allergies against b-lactam antibiotics [124±126]. As a group penicillins had a higher frequency of allergic reactions than cephalosporins. The frequency of reactions has been shown to be greatest with acylaminopenicillins (azlocillin and piperacillin) and imipenem/ cilastatin [127]. The latter combination has not been recommended for routine treatment of CF patients with P. aeruginosa infection [128]. In CF patients with adverse reactions to ceftazidime or other b-lactam antibiotics, a desensitization regimen, starting at a very low dosage and increasing the dosage, has been proposed [129]. Oral antipseudomonal antibiotic therapy in cystic fibrosis For obvious reasons, CF patients prefer oral treatment to intravenous therapy. As a result of the search for a useful oral treatment against P. aeruginosa, the fluoroquinolone ciprofloxacin was introduced in 1985. Early experience has been encouraging [71, 130±132]. Ciprofloxacin resulted in equal improvement in pulmonary function as treatment with azlocillin and gentamicin, and significant improvement was maintained at 6 weeks after ciprofloxacin treatment, which was not observed in a control group [130]. Statistically significant, stepwise changes in clinical scores, pulmonary function, and sputum concentrations of P. aeruginosa were noted, but regression toward initial values occurred by one week after treatment [132]. In a prospective, randomized, double blinded study comparing the efficacy and safety of ciprofloxacin versus ceftazidime/tobramycin in P. aeruginosa infected CF patients it was concluded that sequential intravenous/oral ciprofloxacin monotherapy offers a safe and efficacious alternative to standard parenteral therapy for acute pulmonary exacerbations in CF [133]. Oral ciprofloxacin therapy for 4 weeks has also been shown to prolong the beneficial effects of intravenous antibiotic therapy [134]. However, in seriously affected CF patients conventional treatment was significantly better than quinolone treatment [135]. The use of fluoroquinolones in children and accumulation of data on the pharmacodynamics of these drugs [66] have been limited and delayed by concerns regarding side effects, although adverse reactions were infrequent [132] and no serious side-effects were observed [130]. Animal studies showed that quinolones may negatively influence the cartilage [136] and therefore the use was restricted to patients after puberty. However, ultrasound examination and nuclear magnetic resonance imaging scans showed no evidence of cartilage toxicity in ciprofloxacin-treated CF patients [133, 137]. A comprehensive review of the findings in animals compared with the cumulative published findings in children and adolescents allowed the conclusion that such concerns were not justified [138]. On this topic a consensus meeting has been held [139, 140]. Most experience is obtained with ciprofloxacin, other fluoroquinolones are less well tested in paediatric CF patients with pseudomonal respiratory tract infections. Gastrointestinal side effects, arthralgia in older CF patients with chronic P. aeruginosa infection, and occurrence of sleep disorders and psychotic episodes may be seen as well as photosensitivity reactions to sunshine. Rapid emergence of resistance seems to be a much greater problem than side effects due to cartilage destruction [132]. However, initially, drug resistant organisms were isolated no more frequently after ciprofloxacin than after intravenous therapy with azlocillin and gentamicin. As with other drugs, clinical improvement in CF patients was achieved, regardless of the emergence of resistant P. aeruginosa [130, 132]. However, experience with ciprofloxacin or other quinolones in children <5 yrs of age is very limited. Nebulized antipseudomonal antibiotic therapy in cystic fibrosis In an attempt to circumvent the problem of intravenously administered antibiotics penetrating into lung parenchymal tissue and bronchial secretions and their potential toxicity when given over prolonged periods of time, aerosolization of these drugs has been suggested. Indeed, this approach produced high lung concentrations and low serum levels of antibiotics [47, 141]. However, in general, aerosol delivery is a very inefficient method of delivering the antibiotics and even with the most efficient nebulizer, only a small proportion (~15±20%) of the medication is deposited in the lung; the remainder is either impacted on the oropharynx and swallowed, or exhaled into the surrounding atmosphere [142]. Several studies using aerosolized antibiotics in CF patients have been performed often with variable results [33, 45, 50, 51, 141, 143, 144]. Besides aminoglycosides (gentamicin [144], tobramycin [33, 44, 45, 47, 51] and b-lactam antibiotics (ceftazidime [50] carbenicillin [143]), combinations of both classes have been used [50, 60]. Colistin has also been used [43, 145, 146]. The penicillins have been mostly avoided because of concerns regarding development of hypersensitivity, odour, effect on teeth and greater difficulty in nebulization. Interpretation and comparison of many studies in this field is difficult because of the different designs, the wide range in the amount of the antibiotic used, and the possible carry-over effects in cross-over studies. Despite these methodological difficulties, most of the studies showed

756 G. DOÈ RING ET AL. Lung infection by P. aeruginosa Chronic lung infection by P. aeruginosa Table 2. ± Definitions within the context of management of cystic fibrosis patients Lung colonization Presence of P. aeruginosa in the bronchial tree without direct (inflammation, fever, etc.) or by P. aeruginosa indirect (specific antibody response)signs of infection and tissue damage. Chronic lung colonization Presence of P. aeruginosa in the bronchial tree for at least 6 months, based on at least three P. aeruginosa positive cultures with at least one month in by intervals between them without direct (inflammation, fever etc.) or in direct (specific antibody response) signs of infection and tissue damage. Presence of bacteria in the bronchial tree with direct (inflammation, fever etc.) or indirect (specific antibody response) signs of infection and tissue damage. Infection can also be diagnosed on the basis of a positive antibody response in at least two examinations for patients who do not expectorate and present negative bacterial cultures. Presence of P. aeruginosa in the bronchial tree for at least 6 months, based on at least three positive cultures with at least one month intervals between them with direct (inflammation, fever etc.) or indirect (specific antibody response) signs of infection and tissue damage. Chronic infection can also be diagnosed on the basis of a positive antibody response in at least two examinations for patients who do not expectorate and present negative bacterial cultures. Maintenance antibiotic therapy Prolonged, continuous or intermittent antibiotic therapy, given even if clinical symptoms are discrete and signs of acute exacerbations of lung disease are absent. Prophylactic antibiotic therapy The use of antibiotic before P. aeruginosa has been detected in order to prevent colonization and infection. Antibiotic therapy on demand The use of antibiotics only when symptoms and signs of acute exacerbations of lung disease are present. improved lung function [33, 45, 47, 50, 143, 144] or slowing of the deterioration of lung function [33, 44, 51] in the active treatment group compared to placebo. Additionally, most studies report a reduction in the number of hospital admissions [50, 51, 143, 144] and some report on areductionofp. aeruginosa numbers [47, 145]. Only one study demonstrated a decrease in inflammation by treatment with aerosolized antibiotics [51]. A review of five randomized controlled trials [33, 43, 50, 143] showed benefit for nebulized antipseudomonal antibiotic therapy with no demonstrable adverse effects [147]. A multicentre study including 520 CF patients in the USA demonstrated conclusively that chronic intermittent therapy with tobramycin improved lung function, and reduced hospitalization and concomitant intravenous therapy [33]. Aerosolized colistin in combination with oral ciprofloxacin has been shown to significantly postpone chronic P. aeruginosa infection in CF patients [72, 146] and to maintain pulmonary function during the year after inclusion compared with the control group, in which pulmonary function declined [72]. Similarly, tobramycin inhalation may prevent P. aeruginosa pulmonary infection in CF [148]. Results from a recent consensus conference on the use of aerosolized antibiotics in CF patients [149] and reviews [142, 147, 150] have been published. Due to low serum levels of aerosolized drugs, systemic toxicity is possibly avoided. Nevertheless this depends on the doses administered. Aerosol administration of 400 mg [151] or 600 mg [47] tobramycin performed during tidal volume breathing manoeuvres did not lead to audiometric abnormalities or renal toxicity in CF patients [47, 152]. Similarly, no signs of renal or ototoxicity were detected after long-term tobramycin aerosol therapy [44, 47, 51]. Nevertheless, with higher daily doses of aminoglycosides and other inhalation techniques, long-term safety studies are still required, applying more sensitive methods for the investigation of renal function. Since tobramycin levels in serum samples of CF patients may vary considerably after aerosol treatment [152], general dose recommendations should not be followed but rather serum aminoglycoside concentrations monitored in patients given high doses [64]. Thus, the optimal dose to be aerosolized for the different antibiotics may differ from patient to patient. For colistin, 2 mega units twice daily have been recommended [12]. Monitoring by serum levels is not yet feasible. Dosing of aerosolized antibiotics also depend on clinical efficacy and microbial susceptibility. Emergence of drug-resistance after aerosol therapy has been noticed in several studies [33, 45, 47, 143] although P. aeruginosa susceptibility was regained after drug-free time periods [47]. This led to the concept of intermittent dosing [33, 45]. The effect of resistance on efficacy of more prolonged aerosol antibiotic therapy is unknown at present. Large amounts of the aerosolized antibiotics may also be swallowed and studies have been suggested to investigate the effects on the gastrointestinal flora during chronic administration [64]. Improved drug delivery to the lung may overcome this problem. Caution has to be maintained when solutions of drugs in saline or water are prepared since hypotonic as well as hypertonic solutions may affect lung function leading to bronchoconstriction and inflammation [153±155]. Furthermore, the drug itself can induce bronchoconstriction. Consequently, the use of isotonic solutions and the control of lung function before and immediately after nebulization of the antibiotic as well as therapeutic intervention with bronchodilators has been suggested [12, 64, 155, 156]. For instance, an isotonic solution of colistin is made by dissolving 2 mega units of colistin in 3 ml water and 3 ml physiological saline [12]. Antibiotics are usually aerosolized by nebulizing a solution of the drug, although dry powder inhalation has also been reported [157, 158]. A nebulized aerosol can be produced by an air jet or by ultrasonic sound waves. Nebulization times may be much shorter using ultrasonic devices than using jet nebulizers [64, 115, 159]. Nevertheless, aerosolized medications are still frequently administered by nebulizers. In CF, the nebulizer should produce particles in the range of 2±5 mm in order to reach the smaller bronchioles. Particles too large (>5 mm do not reach the lower respiratory tract, whereas particles too small (<1 mm are exhaled [160]. Important characteristics of nebulizers which may affect performance include drug